2 Oilfield Review

Seismicity in the Oil Field

Vitaly V. Adushkin Vladimir N. Rodionov Sergey Turuntaev Institute of Dynamics of Geospheres, Russian Academy of Sciences Moscow, Russia

Alexander E. Yudin Ministry of Fuel and Energy of the Russian FederationMoscow, Russia

Much of this article originally appeared in the SchlumbergerRussian version of the Oilfield Review, NeftegasovoyeObozreniye 5, no. 1 (Spring 2000): 4-15. For help in prepa-ration of this English version, thanks to David Leslie,Schlumberger Cambridge Research, England; and YefimMogilevsky, Graphics International, Houston, Texas, USA.Results in this article were based on data obtained by thelocal seismic network of Stock Joint Company “Tatneft.”The authors thank I.A. Iskhakov, head of the TNGF seismiccrew, and K.M. Mirzoev, chief of the Tatarstan seismic sur-vey, who provided the catalogue of seismic events and theproduced and injected fluid volumes data. The support from“Tatneft” and the Russian Foundation for Basic Research(RFBR project # 98-05-64547) is gratefully acknowledged.

In some regions, hydrocarbon production can induce seismic activity. To help

understand how production affects seismicity, a recording network was installed

in a producing field in Russia. In a cooperative project between Schlumberger

and the Institute of Dynamics of Geospheres at the Russian Academy of

Sciences, scientists are analyzing the recorded data to help forecast seismic

events, understand reservoir properties and monitor water injection.

Scientists have observed that earthquakes canbe triggered by human action. Induced seismicity,or seismic activity caused directly by humaninvolvement, has been detected as a result ofwater filling large surface reservoirs, develop-ment of mineral, geothermal and hydrocarbonresources, waste injection, underground nuclearexplosions and large-scale construction pro-jects.1 It is important to understand the condi-tions under which seismicity may be induced sothat these operations can be performed safely.

The notion that human activity can provokeearthquakes is not new. In the 1870s, proposalsfor impounding water in man-made lakes acrossregions of southern California, USA, wererejected because of concerns that this might trig-ger earthquakes.2 The hundreds of small earth-quakes detected immediately after the 1936filling of the Hoover Dam in Nevada and Arizona,USA, provided the first definite evidence of suchan effect. Since then, more than 100 other caseshave been reported around the world.3 In someinstances, the resulting seismic activity has beensevere. Within four years of completing construc-tion in 1963, the reservoir area surrounding theKoyna Dam near the west coast of India experi-enced several significant earthquakes, thelargest being a major event of magnitude 7.0.4 In

the nearby town of Koynanagar, masonry build-ings were destroyed and 200 people died.

In the early 1920s, geologists in south Texasnoted faulting, subsidence and earthquakes inthe vicinity of the Goose Creek oil field. Housesshook and faulting broke the earth’s surface.5 Adirect relationship was proposed between oilextraction and the onset of subsidence and seis-mic activity. At the time, subsidence associatedwith hydrocarbon extraction was consideredrare, and this case was thought to be a uniqueoccurrence in geological literature. Similar obser-vations were then reported for the Wilmingtonoil field in Long Beach, California, USA, where sixsmall earthquakes occurred between 1947 and1955, and surface subsidence reached 9 m [30 ft]in 1966 after 30 years of oil production.6

By the 1960s, it became clear that deep injec-tion of fluid could also cause seismicity. Early in1962, waste-water by-products from the RockyMountain Arsenal near Denver, Colorado, USA,were injected into a disposal well in fracturedPrecambrian rocks at a depth of about 12,000 ft[3660 m]. Earthquakes up to magnitude 4.3 beganoccurring one month later, and continued for thethree-year injection period. The frequency ofearthquake occurrence was clearly related to therate and pressure of fluid injection.7

Summer 2000 3

Seismologists speculated that if the physicalbasis for triggering earthquakes by injection couldbe clearly established by field experiments, fluidinjection or extraction might become a means ofcontrolling earthquakes or preventing inadvertentseismic activity. Geophysicists and hydrologistsdesigned an experiment to test the feasibility ofcontrolled earthquake generation in the Rangelyoil field in western Colorado. The field had beenon waterflood since 1957, and an array of seis-mographs in the neighboring state of Utah hadbeen recording small earthquakes in the fieldsince its installation in 1962. In 1967, a portablearray of seismographs was installed directly overthe field. It began recording and locating seismicevents along a subsurface fault in two areaswhere waterflooding had induced high pore

pressures.8 The project successfully initiated seis-mic activity by injecting even more water andhalted seismic activity by producing from near thefault. The report suggested the technique mightbe useful for controlling the timing and size ofmajor earthquakes, and noted that up to thattime, fluid injection for enhancing oil recovery hadnot triggered any damaging earthquakes.

In all these cases, the result of human inter-ference was to change the state of stress in thesurrounding volume of earth. If the stress changeis big enough, it can cause an earthquake, eitherby fracturing the rock mass—in the case of min-ing or underground explosions—or by causingrock to slip along existing zones of weakness.The situation in regions of hydrocarbon recoveryis not always well understood: in some places,

extraction of fluid induces seismicity; in others,injection induces seismicity. In many areas wherethe rock is not under large tectonic stresses, theseismic energy released during induced events islow—typically of magnitude 0 to 3—and noteven felt on the earth’s surface. However, if therock mass is already under large tectonicstresses, the energy added by man’s endeavorscan have a destabilizing influence. Even minoractions can trigger strong seismicity.9

Long-term hydrocarbon exploitation can dis-turb conditions around oil and gas reservoirs inseveral ways, causing significant stress changesin the reservoir and the surrounding rocks.Injected fluid can propagate or filter into cracksand cause increased fluid pressure in pores and fractures, serving as a kind of lubricant in

Induced by dams Induced by oil and gas recovery Induced by mineral-deposit exploitation

1. Nikolaev NI: “On the State of Study of InducedEarthquakes, Related with Industrial Activity” in: An Influence of Industrial Activity on the SeismicRegime. Moscow, Russia: Nauka, 1977 (in Russian).Gupta H and Rastogi B: Dams and Earthquakes. New York,New York, USA: Elsevier Scientific Publishing, 1976. Pasechnik IP: “Earthquakes Induced by UndergroundNuclear Explosions,” in: An Influence of IndustrialActivity on the Seismic Regime. Moscow, Russia: Nauka (1977): 142-152 (in Russian).Simpson DW: “Triggered Earthquakes,” Annual Reviewof Earth and Planetary Science Letters 14 (1986): 21-42. Nicholson C and Wesson RL: “Earthquake HazardAssociated with Deep Well Injection—A Report to theUS Environmental Protection Agency,” US GeologicalBulletin vol. 1951, 1990.

Milne WG and Berry MJ: “Induced Seismicity in Canada,”Engineering Geology 10 (1976): 219-226.Grasso J-R: “Mechanics of Seismic Instabilities Inducedby the Recovery of Hydrocarbons,” Pure and AppliedGeophysics 139, no. 3/4 (1992): 507-534.

2. Bolt B: Earthquakes: A Primer. San Francisco, California, USA: W.H. Freeman and Company, 1978.

3. Guha SK and Patil DN: “Large Water-Reservoir-RelatedInduced Seismicity,” in Knoll P (ed): Induced Seismicity.Rotterdam, The Netherlands: AA Balkema Publishers(1992): 243-266.

4. Earthquake magnitudes in this article are taken from a variety of literature sources. Magnitudes are usuallycalculated from the recorded amplitude of a seismicwave of specified frequency and calibrated for the distance from the earthquake and the magnification of the seismograph.

5. Pratt WE and Johnson DW: “Local Subsidence of the Goose Creek Oil Field,” Journal of Geology 34, no. 7-part 1 (October-November 1926): 577-590.

6. Segall P: “Earthquakes Triggered by Fluid Extraction,”Geology 17, no.1 (October 1989): 942-946.

7. Evans DM: “Man-Made Earthquakes in Denver,”Geotimes 10, no. 9 (May-June 1966): 11-18.

8. Raleigh CB, Healy JH and Bredehoeft JD: “AnExperiment in Earthquake Control at Rangely, Colorado,” Science 191, no. 4233 (March 1976): 1230-1237.

9. Simpson, reference 1.Sadovsky MA, Kocharyan GG and Rodionov VN: “On theMechanics of Block Rock Massif,” Report of the Academyof Sciences of USSR 302, no. 2 (1988): 193-197 (in Russian).Rodionov VN, Sizov IA and Kocharyan GG: “On theModeling of Natural Objects in Geomechanics,”in:Discrete Properties of Geophysical Medium. Moscow,Russia: Nauka (1989): 14-18 (in Russian).

K A Z A K H S T A N

T U R K M E N I S T A N

U Z B E K I S T A N

Gazli

R U S S I A

T U R A NP L A

TE

> Location of the Gazli field, Uzbekistan.

fractured zones. Three types of forces help initi-ate filtration-induced earthquakes as well asother man-made and tectonic earthquakes bycausing motion of rock blocks along faults: First,poroelastic forces can force displacement alonga fault in the surrounding rock mass. Second,hydrostatic forces can transfer pore pressurefrom an injection zone to a zone preparing for anearthquake through a fault or other permeablefeature. Fluid migration in this case may be neg-ligible. Finally, pressure differences can causefluids to migrate from injection zones to zones ofearthquake incipience.

Hydrocarbon field development alwaysinduces at least minor changes in the stress stateof a reservoir. Sometimes this increases the levelof small, background seismic events. The energyreleased depends on the properties of the reser-voir and surrounding rocks, the level of hetero-geneity and the rate at which they weredeposited. Some 40 examples are known in whichreservoir production caused significant changes inthe seismic activity of a neighboring region.Comparison of data from these reservoirs withmeasurements from 200 other fields around theworld shows which properties are most closelyrelated to production-induced seismicity (above).

Average reservoir depth and thickness appear tobe greater for oil fields with induced seismicitythan average depth and thickness values forother hydrocarbon fields. Average porosity andpermeability are lower for hydrocarbon fieldswith induced seismicity than for those without.Initial reservoir pressure has the same distribu-tion in both cases.

Although there are examples of significantearthquakes related to reservoir development,and it is sensible to consider triggered seismicityas one of the possible hazardous consequencesof production, it is rare for reservoir developmentto lead to earthquakes strong enough for peopleto feel. More often, induced seismic events areweak, and can be recorded only with the help ofa sensitive seismometer network.

These feeble seismic events, induced ones aswell as those caused by natural deformation pro-cesses, carry important information about thelocation of zones of weakness and seismicallyactive faults in the rock. They also contain infor-mation about temporal changes in stress state

4 Oilfield Review

> A comparison of probability distribu-tions for some key variables in hydrocar-bon reservoirs. The black line representsdata from 40 oil and gas fields with anobserved increase in seismic activity dueto hydrocarbon production; the red linecorresponds to random-sample data on200 reservoirs located in various regionsaround the world.

110000 0

0

2

4

4

6

8

10

200 400 500300 2 3 4

1

2

0

1

2

3

4

5

20000

400 600 800 1000 1200

1

2

3

Reservoir thickness, m

Prob

abili

ty d

ensi

ty x

10-3

Prob

abili

ty d

ensi

ty x

10-4

Prob

abili

ty d

ensi

ty x

10-2

Prob

abili

ty d

ensi

ty x

10-

3

Initial reservoir pressure, atm

Reservoir depth, km

Permeability, mD Reservoir porosity, %

5

4

3

3

2

1

000 2 4 6 8 10 20 40 60 80

Prob

abili

ty d

ensi

ty x

10-4

Summer 2000 5

and other formation properties. Interpretingrecords of production-induced seismicity allowsidentification of active faults, delineation of fluid-contrast fronts and estimation of time variationsof reservoir permeability and porosity. This infor-mation, in turn, may help to optimize the schedul-ing of hydrocarbon production, water injectionand enhanced recovery operations.

In the following sections, we examine therelationship between recorded seismic eventsand the evolution of hydrocarbon exploitationparameters through two case studies. The first isa study of earthquakes in the region of the Gazligas field in Uzbekistan. The second is an investi-gation of temporal and spatial characteristics ofseismicity in the region of the Romashkino oilfield in Tatarstan, Russia.

Gazli EarthquakesThe Gazli gas field is located in Central Asia about100 km [63 miles] northwest of Bukhara,Uzbekistan (previous page, bottom). The field struc-ture consists of Jurassic, Cretaceous, Paleoceneand Neocene formations overlapping Paleozoicbasement in an asymmetrical anticline with dimen-sions 38 by 12 km [24 by 7.5 miles] (above). Thethickness of sediments is about 1000 m [3300 ft],reaching a total depth of 1600 m [5200 ft].

The field has 11 accumulations—10 gas andcondensate, and one oil—all located in Creta-ceous sediments. Producing horizons consist ofsandstone and clay beds. Porosity of the sandstoneis high and averages 20 to 32%. Permeability of allbut one producing horizons ranges from 675 to1457 mD. Produced gas consists mainly ofmethane (93 to 97%) with condensate in the lowerhorizons (8 to 17.2 g/m3 [67 to 144 lbm/gal]).

The gas field was discovered in 1956 and pro-duction began in 1962. Over the next 14 years,roughly 600×106 m3 of water, or 106 ton per km2,were injected. In spite of the water injection,subsidence was detected at the surface. Thesubsidence rates averaged 10.0 mm/a [2.5 in./yr]in the period 1964 to 1968 and 19.2 mm/a [5 in./yr] from 1968 to 1974. Subsidence wasobserved to be associated with reduction in for-mation pressure: when formation pressuredropped by 1 atm [101 kPa], the central part ofthe field subsided by 2 mm.10

Beginning in 1976, a series of large earth-quakes was recorded. The first significant earth-quake occurred on April 8, 1976 at a distance of

420440

460

400

480500520

-560-580

Section b

Section c

500 Isolines in meters

Wells

Tectonic discontinuities

Gas reservoir contour

Section b

Dept

h, m

Gas

Oil

Water

Clay

North12 11 10 3

400

600

800

1000

Section c

24 15

400

600

800

1000

1200

Dept

h, m

North

Well no. South

Well no. South

1310118

35 23 14

5 km3 miles

5 km3 miles

5 km3 miles

< Structural map (top) and cross sections from the Gazli gas field. The structural map shows well locations, contours of the uppermosthorizon in meters, tectonic disconti-nuities, the locations of the cross sections and the limit of the gasreservoir. The cross sections showgas, oil, water and clay layers.[Adapted from Zhabrev IP (ed): Gas and Condensate Deposits.Moscow, Russia: Nedra, 1984 (in Russian)].

10. Piskulin VA and Raizman AP: “Geodesic Investigations of the Earth Surface Deformation in Epicentral Zones of Gazli Earthquakes in 1976-1984,” Proceedings of 7thInternational Symposium on the Earth Crust ModernMotion. Tallinn, 1986.

20 km [12 miles] from the Gazli gasfield boundary.The earthquake magnitude measured 6.8. Just 39days later, on May 17, 1976, another severeearthquake occurred 27 km [17 miles] to the westof the first one. The magnitude of the secondearthquake was 7.3. Eight years later, on March20, 1984, a third earthquake occurred 15 km [9 miles] to the west of the second earthquake,with a magnitude of 7.2. The hypocentral depthsof all three were 25 to 30 km [16 to 18 miles], allwithin the 32-km [20-mile] thick earth crust in theregion. Aftershocks occurred in a volume sur-rounding the three hypocenters. These earth-quakes are the strongest of all the knownearthquakes in the plain of Central Asia.

There was no clear relationship between thelocation of the earthquake hypocenters and anypreviously known active tectonic structures.Closer investigation showed that the earth-quakes had created new faults.11 Analysis of thefine-scale structure of the aftershock zone indi-cated an initial state of tectonic activation.12 Theorientation of the fault plane, the direction offault-block displacement and the trend of theaftershock zone correspond to the regional stressfield and orientation of regional-scale faults.

Geodesic measurements were made aftereach large earthquake (below). The area that hadpreviously subsided was found to have subsidedan additional 230±8 mm [9 in.] after the 1976earthquakes (above).13 In the vicinity of the earth-

quake epicenters, an upward displacement of thesurface was detected: up to 830 mm [33 in.] near the epicenter of the April 1976 earth-quake, up to 763 mm [30 in.] near the epicenterof the May 1976 earthquake, and up to 751 mm[29.5 in.] near the epicenter of the March 1984earthquake. Horizontal displacements of up to 1 m [3.3 ft] were detected and found to bedirected mainly away from the epicenters.

The amassed data indicate that the Gazliearthquakes were triggered by exploitation of thegas field.14 High tectonic stresses are typical for

border regions of young platforms such as theTuran plate. These stresses cause accumulationof significant tectonic energy. Depletion of thegas field served as a trigger for the release ofaccumulated tectonic energy in the form of sig-nificant seismic events. Field production wasundertaken without consideration of the possibil-ity of production-induced seismicity. Some geo-physicists, including the authors of this article,believe that if the natural tectonic regime hadbeen taken into account during the planning ofhydrocarbon recovery, the earthquakes mighthave been avoided.15

6 Oilfield Review

-100

+800+500 +600

+2000

+751

-1000

+100

+400

0

Gazli

+830

+200

0

+100

+751

0

Chorbakty

Karakyr

Vertical displacement after the earthquakes in 1976, in mmVertical displacement after the earthquakes in 1984, in mmA boundary of gas accumulationEpicenters of the earthquakes on April 8 and May 17, 1976 and March 20, 1984Tectonic faults

+100+300

-100

Position of surface vertical displacement profile

+200

10 km6 miles

< Surface deformation following the Gazliearthquakes of 1976 and 1984. Maximumvertical displacements are shown asblack dots, and earthquake epicenters asred dots. Dashed lines mark the verticaldisplacement following the 1976 events;and solid lines mark the vertical displace-ment following the 1984 event. The gasfield is shaded in light red. Tectonic faultsare shown as thick black lines. The redvertical line marks the position of thecross section displayed (above). [Adaptedfrom Piskulin and Raizman, reference 10].

0

600

400

200

-200

South

Disp

lace

men

t, m

m

0 10 20 30 40

North

Gazli fault Karakyrfault

Distance, km

1964 to 19681968 to 19741974 to 1976

Measurement periods

800

> North-south profile of vertical displacement after the Gazli earth-quakes. The region between the Gazli fault and the Karakyr faultsubsided, while north of the Karakyr fault, upward displacementwas measured.

Summer 2000 7

The Romashkino Oil Field The Romashkino oil field is the biggest oil field in Russia (right). It has a maximum dimension ofabout 70 km [44 miles], a structural height of 50to 60 m [164 to 197 ft] and a reservoir depth of1600 to 1800 m [5200 to 5900 ft].16 The deposit isa succession of 10- to 30-m [33- to 100-ft] thickoil-bearing Devonian sandstones and carbonaterocks (below). The main productive formationcontains thinly bedded sandstones and clays.Permeability of the sandstone layers is 200 to420 mD, porosity is 18.8 to 20.4% and oil satura-tion is 69.4 to 90.5%. Initial reservoir pressurewas 160 to 180 atm [16.2 to 18.2 MPa].

Geological exploration in this region began in1933. In 1947, exploration drilling commenced,and in 1948 Romashkino produced its first oil.17

Water injection began in 1954, but for the firstseveral years, injection did not compensate forfluid extraction. In 1958, for the first time, the vol-ume of fluid injected that year exceeded the vol-ume of fluid extracted, and by 1963 total injectedand extracted fluid volumes balanced. By 1975,the total volume of fluid injected in the programreached 2.13×109 m3, or 104.7% of totalextracted fluid. Suggested maximum pressuresfor water injection were 200 to 250 atm [20.2 to25.3 MPa], but actual injection pressures some-times were higher.

11. Shteinberg VV, Grajzer VM and Ivanova TG: “GazliEarthquake on May 17, 1976,” Physics of the Earth 3(1980): 3-12 (in Russian).

12. Turuntaev SB and Gorbunova IV: “CharacteristicFeatures of Multi Fracturing in Epicentral Zone of Gazli Earthquakes,” Physics of the Earth 6 (1989): 72-78 (in Russian).

13. Piskulin and Raizman, reference 10.14. Akramhodzhaev AM and Sitdikov B: “Induced Nature

of Gazli Earthquakes, a Forecast of Earthquakes of Gazli Type and Prevention Measures,” Proceedings ofWorkshop on Experience of Gazli Earthquakes StudyFurther Investigation Directions. Tashkent, Uzbekistan:FAN (1985): 59-60 (in Russian).Akramhodzhaev AM, Sitdikov BB and Begmetov EY:“About Induced Nature of Gazli Earthquakes inUzbekistan,” Geological Journal of Uzbekistan 4 (1984):17-19 (in Russian).Volejsho VO: “Conditions of Gazli EarthquakesOccurrence,” Proceedings of Workshop on Experienceof Gazli Earthquakes Study Further InvestigationDirections. Tashkent, Uzbekistan: FAN (1985): 65-66 (in Russian).Mavlyanov GA (ed): Gasli Earthquakes in 1976 and 1984.Tashkent, Uzbekistan: FAN, 1986 (in Russian).

15. Akramhodzhaev et al, reference 14.16. Bakirov AA (ed): Geological Conditions of Oil and Gas

Accumulation and Location. Moscow, Russia: Nedra,1982 (in Russian).

17. Muslimov RH: An Influence of Geological StructuresDistinguish Features on an Efficiency of Romashkino Oil Field Development. Kazan, Russia: KSU, 1979 (in Russian).

K A Z A K H S T A N

T U R K M E N I S T A N

U Z B E K I S T A N

Romashkino

R U S S I A

> Location of the Romashkino field, Russia.

1400

1450

1500

1550

1600

1650

Limestone

Clay

Oil reservoir

Basement

140 4811-88 518 519 14-91 27 33 30 627 19-553 18-552 16-551 8-550Wells

Sandstone

NW SE

Dept

h, m

10 km

6 miles

> Geologic profile of the Romashkino field. [Adapted from Muslimov, reference 17].

For exploitation convenience, the Romashkinooil field is divided into more than 20 areas. Inthese areas, various methods of injection areused: injection through a line of wells, localinjection wells and pattern waterflooding. In sev-eral areas, well density is three to five wells perkm2. However, overall, the density of well cover-age and geometry of well location appear to bethe result of a complicated development historydefined by objective factors as well as randomones.18 Methods of nonstationary injection wereused in a number of areas. That is, water wasinjected through one injection line for one month,then the first line was switched off and waterwas injected through another line, and so on.Injected water migration velocity varies from 100 to 1500 m/a [330 to 4900 ft/yr].19

Characteristics of Romashkino Seismicity According to seismic zoning maps, the southeastpart of Tatarstan in the region of the Romashkinooil field is considered a seismically quiescentarea. But in 1982 and 1983, after decades ofproduction and injection, citizens in the vicinity ofthe town of Almetjevsk began noticing moderateseismic events. In 1985, the “Tatneftegeophysica”seismic service installed a local seismic networkthat recorded numerous earthquake epicenters inthe region of the Romashkino oil field (left). Mostof these are in the western part of the field on theAltunino-Shunaksky depression, the structuralboundary between the Romashkino and Novo-Elkhovskoye oil fields.

From 1986 to 1992, the network recorded 391local earthquakes with magnitudes up to 4.0.Three time intervals showed noticeable increasesin seismic activity—at the end of 1986, in themiddle of 1988 and at the end of 1991. The largestepisodes were an earthquake on September 23,1986, with magnitude 3.8 and another with mag-nitude 4.0 on October 28, 1991, in the region ofthe town of Almetjevsk.20

8 Oilfield Review

10 km6 miles

Romashkino oil field

Almetjevsk1

2

3

Leninogorsk

Novo-Elkhovskoye oil field4

5

Seismic recording stationsAlmetyevneft producing areasBerezovskaja (B area)Severo-Almetyevskaja (S area)Almetyevskaja (A area)Minibayevskaja (M area)

Limits of the Romashkino andNovo-Elkhovskoye oil fieldsIsointensity lines of the Sept. 23, 1986earthquakeCross section

Legend

3

Altun

ino-S

huna

ksky

depre

ssion

struc

ture

Energy classification of seismic events5 10

> Seismic activity in the region of the Romashkino oil field.Seismic recording stations are triangles, seismic epicentersare dots or circles, with the size depending on the amount of energy released. The dashed red ellipses are contours ofintensity of the September 1986 earthquake. The black linepositions the cross section displayed (previous page, bottom).The four producing areas of the Romashkino field that exhibit the most seismicity are delineated (B, S, A and M) and discussed in later sections of the article. [Adapted fromIskhakov et al, reference 20.]

N

1

50

100

500

1000

2000

Altu

nino

-Shu

naks

kyde

pres

sion

stru

ctur

e

J /km21/3

Quantifiedseismicactivity,

10 km6 miles

> Distribution of quantified seismic activity (color coded) in theregion of the Romashkino oil field. Quantified seismic activity is the sum of the cube root of seismic-event energy occurring in onesquare km2. The distribution of seismic activity is related to tectonicfaults (solid purple lines) and the Altunino-Shunaksky depressionstructure (dashed lines).

Summer 2000 9

The recorded activity can be examined in sev-eral ways to compare it to reservoir parameters.A map of seismic activity in the region of theRomashkino oil field shows spatial variations in the level of activity (previous page, bottom). A quantitative measure of seismic activity wascomputed for each km2 by summing the cuberoots of the energies in all earthquakes occurringthere during the 1986 to 1992 period of observa-tions.21 Most of the seismic activity quantified in this way is situated along the Altunino-Shunaksky depression, with some correspondingto mapped tectonic faults.22

Before the recorded seismic activity can beused more quantitatively, the quality of the datamust be assessed. Seismic recording networkshave sensitivity limitations in the magnitude anddistance of events they can record. Extremelysmall events can go undetected, as can distantevents. Also, since large events do not occuroften, shorter seismic recording intervals are lesslikely to record the larger earthquakes. For allearthquakes in a given region, a linear relation-ship exists between the magnitude of seismicevents recorded in a time interval and the loga-rithm of the number, or frequency, of events ofthat magnitude. If the frequency-magnitude plotshows deviations from a linear trend, the earth-quakes being plotted are not representative of allthe seismic activity in the region. A deviationfrom linear on the low-magnitude end indicatesthat the seismic network is not sensitive enoughto weak events, while a deviation on the high-magnitude end usually shows that the observa-tion period was not long enough.

In the case of the seismic activity recordedfrom the Romashkino network, the frequency-magnitude plot is mostly linear (above). Onlythose events that were listed in the 1986 to 1995catalogs of instrumentally recorded seismicevents were plotted. Remote events with epicen-tral distances of more than 70 km [44 miles] werenot considered. During the observation period,

different catalogs used different methods of seis-mogram interpretation. To ensure consistency,frequency-magnitude relations were plottedseparately for three different time intervals: 1986through 1987, 1988 through 1992, and 1992through 1995. Also, an average annual number ofevents for these time periods was considered.

For the earlier time interval—up to 1987—only the events with magnitude greater than 2 are representative for this particular seismicnetwork: not enough events with lower magni-tude were recorded. After 1987, because of achange in the seismic network, events with mag-nitude 1.5 become representative, and so can beincluded in further calculations.

For all three time intervals, the slopes of thefrequency-magnitude plots range from –1.02 to–1.3, considerably more negative than the valuefor natural seismicity, which is –0.75 to –0.9.23

The slopes of the Romashkino plots reach valuestypical of induced and triggered seismicity, asmeasured elsewhere in the world.24

Change in Quantified Seismic Activity with TimeQuantified seismic activity is one of the mostuseful parameters for characterizing seismicity.25

It provides a way to transform the display of seis-mic events from a discrete system to a continu-ous one: the point-by-point representation ofseismic events described by three spatial coordi-nates plus the event time and energy converts toa continuous plot in a different coordinate sys-tem. The selected quantitative measure of activ-ity was described earlier as the sum of the cuberoots of the energies in all events occurring in akm2. To minimize the influence of an arbitrarychoice in the way the area is divided into squaresand in the selection of a beginning time interval,activity values were computed for overlappingareas and time intervals. The amount of overlapdepends on the smoothness of the obtained dis-tributions of activity.

18. Muslimov, reference 17.19. Sultanov SA: A Control of Water Injection in Oil

Reservoirs. Moscow, Russia: Nedra, 1974 (in Russian).20. Iskhakov IA, Sergeev NS and Bulgakov VYu, A Study of

Relation Between Seismicity and Oil Fields Development.A report of OMP 50/81. Bugulma, Russia:Tatneftegeophysica, 1992 (in Russian).

21. Energy is calculated through a formula based on thesquare of the amplitude of seismic waves of specifiedfrequency content measured at a standard distancefrom the event source.

< Relationship between the logarithm of the number of seismic events and the magni-tude of the events in the regionof the Romashkino oil field.

Nurec Dam,” in Seismological Investigations in theRegions of Large Dam Constructions in Tajikstan.Dushanbe, Tajikstan: Donish (1987): 101-119 (in Russian).Turuntaev SB: “An Application of the Weak SeismicityCatalog for Detection of Active Faults in Rock Massif,” in The Monitoring of Rock Massif State During Long-Time Exploitation of Large-Scale Underground Works.Appatity, Russia: KFAS, 1993 (in Russian).

25. Ponomaryov VS and Tejtelbaum UM: “DynamicsInteractions Between Earthquakes Focuses,” in:Regional Investigations of Seismic Regime. Kishinev,Moldova: Shtinitsa (1974): 79-92 (in Russian).

22. Belousov TP, Muhamediev ShA, Turuntaev SB, Junga SL,Ischakov IA and Karakin AV: “Active Faults, StressesState and Seismicity of South-East Tatarstan,” Seismicityand Seismic Zones of Northern Part of Eurasia, part 2.Moscow, Russia: Nauka (1994): 90-108 (in Russian).

23. Sadovsky MA and Pissarenko VF: Seismic Process inBlock Media. Moscow, Russia: Nauka, 1991 (in Russian).Isacks B and Oliver J: “Seismic Waves with Frequenciesfrom 1 to 100 Cycles Per Second Recorded in a DeepMine in Northern New Jersey,” Bulletin of the Seismo-logical Society of America 54, no. 6 (1964): 1941-1979.

24. Mirzoev KM, Negmatullaev SH and Dastury TYu: “AnInfluence of Mechanical Vibrations on CharacteristicFeatures of Seismic Energy Release in the Region of

Num

ber o

f sei

smic

eve

nts,

N

0.1

1

10

100

1.0 1.5 2.0 2.5 3.0 3.5 4.0Magnitude, M

1986 to 1987

1988 to 1992

1992 to 1995

Initially, the temporal and spatial componentsof activity change were calculated separately.The variation over time was examined on amonthly basis by summing the cube roots ofenergies of events that took place during amonth. The resulting temporal series was nor-malized by the average value for that time period(above, top). Two strong peaks and severalsmaller ones are evident in this plot, but period-icity, if it exists, is not obvious. The seismic activ-ity also may be displayed in other ways to try toextract any underlying periodicity. These meth-ods involve transformation to phase coordinates(see “Another Dimension in Seismic Activity,”page 12 ). Looking at the data in the new coordi-nate system led to the following results.

Over the observation period, seismic activityin the Romashkino oil field occurs in two cycles.Both cycles start with the strongest earthquakesfor this region and each cycle lasts for about five years. The two cycles of activity variations

from 1986 to 1990 and from 1991 to 1995 can be smoothed and superimposed so that their first maximums coincide (above, near). An intrigu-ing qualitative agreement of the curves appears,presenting evidence of some kind of regularity inseismic activity oscillations.

The existence of a regular component to thesequence of seismic events carries informationabout the energy state of the rock. It seems pos-sible that when the level of energy accumulatedin the rock from both natural and human sourcesreaches a certain value, energy is released byseismic events that are structured in space andtime. This is similar to the behavior of a fluidbeing heated: for certain values of the rate ofenergy supplied to the fluid, its laminar move-ment changes to a chaotic flow, and then to aregular flow with convection cells.

In a rock formation undergoing oil productionand water injection, there is an increased possi-bility of a large earthquake, regardless of therelease of natural tectonic deformation energy inthe form of seismic events. This is because theenergy transferred to the rock through hydro-carbon exploitation will continue to increase. The existence of quasi-periodic oscillations in the level of seismic activity suggests that theinput energy is rather large. Understanding this relationship between seismicity andexploitation regimes may allow seismicity to becontrolled by a more careful scheduling of pro-duction and injection.

Spatial Characteristics of Romashkino SeismicityThe seismic behavior of the Romashkino oil fieldexhibits an interesting characteristic: a high num-ber of earthquakes occur in pairs, with a shorttime between members of a pair. For example,about 60 paired events with magnitude less than1.0, or about 50% of the total number of eventswith such magnitude, occurred within 24 hours ofone another. One can suppose that eventsgrouped in time are somehow also connected inspace. Examples of this can be seen in laboratorystudies of seismic signals generated during crackgrowth in block models of rock.26 Under certainconditions of crack development, a seismicimpulse is generated at the moment a crackreaches the block boundaries. The locations ofthe event pair define the limits of the episodicmovement along the crack, or fault.

Connections between epicenter pairs in theRomashkino field generally show north-southalignment, trending with the longitudinalAltunino-Shunaksky depression (next page, top).This direction also corresponds to the model ofthe regional stress field.27

Correlating Seismic Activity withHydrocarbon ExploitationIt is always difficult to know whether seismicityis the result of human modifications in the regionor if it is natural seismic activity related to tec-tonic processes; timing could be the key to know-ing the difference. In general, the answer mightbe obtained if a regional seismic network hadbeen installed in advance of the hydrocarbondevelopment, dam construction or mining opera-tion. The seismic network could record a back-ground level of natural seismicity and quantify its

10 Oilfield Review

26. Turuntaev SB: “A Study of Various Model WavesSources in Application to Seismology,” PhD Thesis,Institute of the Physics of the Earth, Moscow, Russia,1985 (in Russian).

27. Belousov et al, reference 22. 28. Mirzoev et al, reference 24.

> Seismic activity variation in the Romashkino region. Theamplitude was calculated by summing the cube root of energiesfrom all events occurring in that month.

> Comparison of two cycles of seismic activity in the region of the Romashkinooil field.

19860

3

6

9

Year

Seis

mic

act

ivity

, nor

mal

ized

1988 1990 1992 1994 1996

0

1

2

3

4

0 12 24 36 48 60Time, months

Seis

mic

act

ivity

, nor

mal

ized 1991 to 1995

1986 to 1990

Summer 2000 11

characteristics. If, after the beginning of humanaction, a significant change in seismicity charac-ter is recorded, it could reasonably be interpretedas a seismic reaction of the rock formation toman’s intervention.

Installation of seismic recording networks andassessment of background seismic activity arealready common practice in regions where thelevel of natural seismicity is high.28 However, instable areas without a history of natural seismic-ity and where no sizeable earthquakes areexpected, an advance seismic background studyusually is not performed. In the absence of anadvance study, the question may be resolved bytwo methods: first, to compare characteristics ofthe observed seismicity with those of known nat-ural seismicity and those of induced seismic activ-ity; and second, to look for correlation betweenthe natural seismic and human activities.

In the first method, as was shown above, theslope of the magnitude-frequency plot for seis-micity in the region of the Romashkino oil fieldhas a value more typical of induced than naturalseismicity. But the low number of recordedevents indicates this result may not have highstatistical significance.

The second method involves comparing therecorded seismicity with the exploitation sched-ule of the Romashkino oil field. The relevant pro-duction data are the values of the monthlyvolumes of fluid extracted and injected from 1981

to 1992 for the four most seismically active areasof the Romashkino oil field: Almetyevskaja (A),Severo-Almetyevskaja (S), Minibayevskaja (M)and Berezovskaja (B).

With these values, a pseudocatalog was con-structed to tabulate the monthly extracted andinjected volumes and the volume imbalance, orthe difference between the volumes of injectedand extracted fluids. These values were assigneda date (middle of a month), time (middle of a day),coordinates (approximate center of the consid-ered area), and depth (1 km). Arranged in this for-mat, the production data closely resembled thestandard form for seismic catalogs, but listedfluid volume instead of seismic energy.

The previously described procedure for calcu-lating quantified seismic activity was applied tothe volumes in the pseudocatalog, but this time a“quantified exploitation activity” was calculated(below). The quantified exploitation activity wasalso analyzed using a 6-month moving average:6-month averaged values of extraction, injectionand imbalance were calculated, then the intervalwas shifted by a month and calculated again. Theresults were normalized by the overall average.

N

1

50

100

500

1000

2000

J /km21/3

Quantifiedseismicactivity,

Altu

nino

-Shu

naks

kyde

pres

sion

stru

ctur

e

Azimuth, degrees

Num

ber o

f eve

nt p

airs

12

10

8

6

4

2

0-60 -30 60 90-90 300

10 km6 miles

> Connections between pairs of Romashkinoseismic events. Black lines connect event pairs,purple lines are mapped faults. The insertshows azimuthal distribution of connectionsbetween pairs of events.

(continued on page 14)

Seis

mic

act

ivity

, nor

mal

ized

3

2

1

0

-1

Prod

uctio

n, in

ject

ion

and

imba

lanc

e, n

orm

alize

d 3

2

1

0

9

8

7

6

5

4

3

2

1

0

4

3

2

1

0

Year

ProductionInjectionImbalanceSeismic activity

Year

ProductionInjectionImbalanceSeismic activity

Seis

mic

act

ivity

, nor

mal

ized

1981 1983 1985 1987 1989 1991 1993

1981 1983 1985 1987 1989 1991 1993

Prod

uctio

n, in

ject

ion

and

imba

lanc

e, n

orm

alize

d

> Top: Comparison of monthly values of seismic activity (red) withvariations in total volumes of injection, production and imbalance inthe four central areas (combined) of the Romashkino field. Bottom:Comparison of smoothed values of seismic activity (red) with varia-tions in volumes of injection, production and imbalance in the fourcentral areas of the Romashkino field.

12 Oilfield Review

For many natural processes, periodicity is evi-dent from a simple plot of observation versustime. For example, the periodicities of oceantides, phases of the moon, earth-surface tem-perature, hours of daylight and several otherphenomena are easily recognized from observa-tions or simple plots.

However, some processes may have so manyforces at work that periodicity is not obvious.One way to analyze a time-varying observationcalled A(t) is to write it as a sum of threecomponents

A(t) = Ap (t) + Ar (t) + At (t)where Ap describes the high-frequency randomoscillations of the activity, Ar is the regularcomponent, and At represents slow variations,or a trend.

To find a regular component in the behaviorof the function A(t), we can change coordinatesfrom A(t) and t to phase coordinates A(t) andits derivative, dA(t)/dt. The new coordinatescan be thought of as the activity and the rate of activity variation.

For the seismic example, a point in the new,phase-coordinate system defines a state of theseismic process at some instant of time and thevelocity of change of this state. A set of points,or a trajectory, defines a change of the system with time.

It is known that if a system’s behavior can be described with certain types of equations,then special points, lines and areas in phasecoordinates exist that “attract” the neighboringtrajectories. These points, lines and areas arecalled “attractors.”1

If the system is one of a monotonous decrease,the corresponding attractor is known as a node(below). For any starting time, the systemmoves in a direct line toward that node in phasespace. In a system of damped oscillations, theattractor is known as a focus, toward which thesystem will move. A system of decreasing orincreasing oscillations will have a correspond-ing, elliptically shaped, limit-cycle attractor inphase space. Highly irregular oscillations canstill exhibit some regularity in phase space andbe drawn to multiple attractors.

When there is a change in the parametersdefining the system evolution, the set of possi-ble solutions of the corresponding equations can change too. That may result in a change in the types of attractors in phase space. Such a change in attractor type is called bifurcation.The simplest examples of bifurcation are fromone node (or focus) to two nodes (or foci), abifurcation from focus to limit cycle, or bifurca-tion from one limit cycle to two limit cycles.

Expressing seismic activity in terms of phasecoordinates is useful for several purposes:• Two basic characteristics of the seismic pro-

cess (its activity and rate of activity variation)are considered and transformed as indepen-dent values.

• The resulting phase portrait, or map, is moresensitive to procedures like smoothing andtrend removal, which simplifies the selectionof a time period for the calculation of activ-ity, a smoothing type, or further coordinatetransformation.

Another Dimension in Seismic Activity

A(t)

dA/dta

tA(t) A(t)

dA/dtb

tA(t)

A(t)

dA/dtd

tA(t)A(t)

dA/dtc

tA(t)

> Time-varying functions (left member of each pair) and corresponding types of attractors (right memberof each pair) in phase space. a) node; b) focus; c) limit cycle; d) limit cycle with multiple attractors.

Summer 2000 13

• The standard procedure of Fourier analysis is not effective if applied to quasi-harmonicoscillations with changing frequency andamplitude. In phase coordinates, such changescan still be analyzed in terms of attractors.For example, an increase in the amplitude of oscillations up to a constant value will looklike a growing limit cycle and a decrease inamplitude to zero will look like a point-typeattractor.

• After transforming the phase portrait to aform that allows a mathematical description,one can carry out the reverse transformationand obtain a mathematical description of theregular component of the original seismicprocess. This may allow estimation of futureseismicity. The statistical significance of thisprognosis depends on the value of the random,or unpredictable, component of seismic activ-ity and of rate of activity variation, and it alsodepends on the ability to recognize bifurcationpoints in phase trajectories (points of changeof seismic regime type).

Phase Characteristics of Romashkino Seismic ActivityThe time variation of quantified seismic activityin the region of the Romashkino oil field (left,part a) can be described by a phase portrait(left, part b). At first glance, the activity-statetrajectory in phase coordinates looks chaotic.However, the random component can be removedby moving-window smoothing and the trend canbe removed by a linear transformation similar toaxes shift and rotation (left, part c).

The resulting phase trajectory (left, part d)starts at an initial point then spirals in; at acertain moment the trajectory comes back tothe outer part of the spiral and then spins inagain. All the while, the trajectory remainswithin a certain area.

This phase portrait resembles the limit cycledisplayed on the previous page, part c for anoscillator under the action of an external force.

An outward motion of a spiral trajectory gen-erally corresponds to an increase in amplitudeof seismic-activity oscillations, while an inwardmotion corresponds to a decrease of activityoscillations. Shape and dimensions of theobtained cycles can yield additional informationabout the seismic process, and should be stud-ied further. One observation already is thatoscillations of the seismic activity are notstrictly sinusoidal; the period tends to oscillatewith an average value close to 12 months.

1. Haken H: Advanced Synergetics. Instability Hierarchiesof Self-Organizing Systems and Devices, Springer seriesin Synergetics. Vol 20. New York, New York, USA:Springer-Verlag, 1983.

1986Year

1988 1990 1992 1994 1996Year

0

3

6

9

Seis

mic

act

ivity

, nor

mal

ized

Seismic activity, normalized

dA/d

t

dA/d

t

Seismic activity, normalized

Seis

mic

act

ivity

, nor

mal

ized

0

0

1

1

-1

-1

-1

-2

-4

0

0

1 2

2

4

3 4 5 6 7 8 9 2-2

0

-1

-2

1

2

d

ca

b

1986 to 19901991 to 1995

1986 1988 1990 1992 1994 1996

> Seismic activity variations in the Romashkino region. Amplitude of seismic activity (top left)was calculated by summing the cube root of energies from all events occurring in that month.These data were displayed in phase coordinates (bottom left) to see if periodicity could beidentified. The seismic activity data were then smoothed and detrended (top right) to extract a regular component. The phase portrait (bottom right) of the smoothed, detrended regularcomponent shows some similarities to the phase portrait of a limit cycle on previous page, part c.

An injection effectiveness, or ratio of pro-duced fluid to injected water volumes, was cal-culated for the four most seismically active areas(right). Comparing these to the quantified seismicactivity in the region of the Romashkino oil fieldshows an inverse relationship in the oscillationsof seismic activity and effectiveness of injection.In 1986, the time at which the seismic activitydata become available and show a marked dropfrom extremely high to low, the character of timevariation of the production parameters changesconsiderably.

In Area A, injection effectiveness begins tooscillate with significant amplitude opposite to theoscillations of seismic activity. In Area S, the onsetof injection-effectiveness oscillations is alsoobserved, but they are less synchronized with theseismic-activity oscillations. In Area B, evenclearer, quasi-harmonic, injection-effectivenessoscillations are observed with a period close to 12 months and a regular amplitude. In Area M, a trend of decreasing injection effectivenesschanges in 1986 to an increase with oscillations,roughly opposite in sign to the oscillations ofseismic activity.

To some extent, the features observed in thetemporal variations of injection effectiveness arerelated to a change to a new fluid-injection tech-nology in 1986. One of the results of such achange was a decrease of the injected volume insummer. In winter, injection was maintained toavoid freezing in the flow lines. This introduced aseasonal component to the effectiveness oscilla-tions and more economical water injection ingeneral. At the same time, it is impossible toassert that all the variations are due to injection-technology differences.

The injection effectiveness for the A, S, M andB areas can be compared with the variation inquantified seismic activity in each area (left). Forcompleteness, the seismic activity changes arealso compared with the extracted and injectedfluid volumes.

A notable feature is the increase of injectionvolumes that took place four months before thetwo most considerable increases in seismicity—at the beginning and at the end of the studiedtime period in Area A. It is also remarkable that production decreased in these periods ofincreasing seismic activity, even when injection

14 Oilfield Review

Injection effectiveness Injection effectiveness

Injection effectiveness

0.8

0.9

1.0

1.1

1.2

0

1

2

3

4

Injection effectiveness Seismic activity

0.6

0.8

1.0

1.2

0

1

2

3

4

0.7

0.8

0.9

1.0

0

1

2

3

4

0.6

0.8

1.0

1.2

1.4

0

1

2

3

4

Seismic activity

Berezovskaja

Severo-Almetyevskaja Almetyevskaja

Minibayevskaja

Seismic activity Seismic activity

YearYear

YearYear1981 1983 1985 1987 1989 1991 1993 1981 1983 1985 1987 1989 1991 1993

1981 1983 1985 1987 1989 1991 19931981 1983 1985 1987 1989 1991 1993

> Comparison of overall Romashkino seismic activity with variations of injection effectiveness forthe four different areas of the field. In each case, the scale of the left vertical axis is injectioneffectiveness and the scale of the right vertical axis is normalized seismic activity.

Year1986 1988 1990 1992

0

2

4

6

8

10

12

14

16

18

20

Seis

mic

act

ivity

, nor

mal

ized

Ratio

of v

olum

e of

pro

duce

d flu

id to

vol

ume

of in

ject

ed w

ater

Imba

lanc

e x

103

m3

Volu

mes

of p

rodu

ctio

n an

d in

ject

ion

norm

alize

d on

ave

rage

val

ues

0

0.2

0.4

0.8

1.0

1.2

1.4

0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

-200

200

0

400

600

800

1000

0.6

Year1986 1988 1990 1992

Injectioneffectiveness

AM

BS

Seismic activityAM

BS

ImbalanceAM

BS

ProductionAM

BS

InjectionAM

BS

> Comparison of seismic activity variations (bottom left) with injection effectiveness(top left), production and injection (top right) and volume imbalance (bottom right)for the four most seismically active areas of the Romashkino field.

Summer 2000 15

increased. Later on, even weaker increases of theseismic activity are always accompanied by adecrease in total fluid production in Area A. It isalso interesting that, for example, during the 1991to 1992 increase of seismic activity in Area M,both injection and production increase, but injec-tion effectiveness decreases at the same time.

Regression analysis shows a statistically sig-nificant relationship between the seismic-activityvariations in the four studied areas and produc-tion and injection regimes for these areas. Theconfidence level of the relationship is 99%.

Crosscorrelation coefficients were computedto help understand the relation between seismicactivity in the four most seismically active areasof the Romashkino oil field and some character-istics describing the exploitation process, such asextracted and injected fluid volumes, imbalanceand injection effectiveness. During the period ofstudy, the volume of injected fluids and the vol-ume of produced fluids both decreased for eco-nomic reasons. For completeness, correlationswere computed between the seismic activity anddetrended values—removing a linear trend fromthe values—of injected and produced volumes.

Correlation between the exploitation parame-ters in one area and the seismic activities in allfour areas can be shown graphically (right). Thecorrelation with the seismic activity of eachregion is depicted as a horizontal bar. Longer barsindicate better correlation, and bars to the leftshow negative correlation.

It is remarkable how well the seismic activ-ity and exploitation parameters correlate, notonly within an area but also between areas. Theinjected and produced volumes in every areaand their detrended counterparts correlate pos-itively with the seismic activity in all four areas,with few exceptions (correlations between seis-mic activity in Area A and production in everyarea are negative). The injection effectiveness(produced/injected) in every area correlatesnegatively with seismic activity in all four areaswhile imbalances correlate positively. The high-est absolute values of correlation are observedbetween the detrended production in Area Aand the seismic activity in Areas A and M(which is near A); and between the imbalance inArea M and the seismic activity in Areas S andB. Absolute values of these correlation coeffi-cients are greater than 0.7.

Imbalance

Injection detrended

Production detrended

Production/injectiondetrended

Production/injection

Production

Injection

-0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Severo-Almetyevskaja

Imbalance

Injection detrended

Production detrended

Production/injectiondetrended

Production/injection

Production

InjectionAlmetyevskaja

Imbalance

Injection detrended

Production detrended

Production/injectiondetrended

Production

Injection

Production/injection

Berezovskaja

Activity of all areas

Imbalance

Injection detrended

Production detrended

Production/injectiondetrended

Production

Injection

Production/injection

Severo-Almetyevskaja

Almetyevskaja

Berezovskaja

Minibayevskaja

Correlation-0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Minibayevskaja

> Correlation between exploitation parameters and seismic activity forthe four production sectors. The exploitation parameters are listed at theright. The correlation with seismic activity in each of the four areas isshown as a colored horizontal bar. For example, at the top, the correlationcoefficient between the production/injection ratio in the S sector and theseismic activity in the A sector (blue bar) is –0.12.

The correlation between seismic activity andhydrocarbon exploitation means the two arerelated, but it does not indicate which one is thecause, which one is the effect, and how long ittakes the cause to create the effect. Shifting thedata series in time relative to each other, recom-puting the correlation, and tracking the lag thatresults in the best correlation gives the best sta-tistical estimate of the time lag between causeand effect (right and next page). Positive lags cor-respond to positive time shifts of the seismic-activity series relative to other data series. Themost interesting are the plots for Areas M and A,which indicate that changes in exploitation param-eters precede changes in seismic activity. Forthese areas, maximum correlation is observedwhen lags are positive and equal to one to twomonths. Correlation coefficients reach 0.8 forArea M (correlation between seismic activity andinjection) and 0.7 for Area A (correlation betweenseismic activity and imbalance, and between seis-mic activity and production).

The maximum correlation for Area B corre-sponds to zero or negligible time shift.

It was a surprise that for Area S, the maxi-mum correlation occurs when the time shifts ofseismic activity relative to most parameters arenegative and equal to six to seven months. Thismeans that the change in the seismic activityprecedes the change of exploitation parameters.

Exploiting SeismicityFew will deny that there is a relationshipbetween hydrocarbon recovery and seismic activ-ity, but exactly how strong a relationship existshas yet to be determined. Furthermore, what canor should be done about it sparks another debate.

In regions of high tectonic potential energy,hydrocarbon production can cause severeincreases in seismic activity and trigger strongearthquakes, as in Gazli, Uzbekistan. In regions oflower tectonic stress, earthquakes of that magni-tude are less likely, but relatively weak earth-quakes could occur and damage surface structures.

16 Oilfield Review

Activity with Injection (detrended)

Lag, months-25 -15 -5 5 15 25

-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

Activity with Production (detrended)

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

Activity with Injection Effectiveness

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

Activity with Imbalance

-25 -15 -5 5 15 25

25

-1.0

-0.6

-0.2

0.2

0.6

1.0Cr

ossc

orre

latio

ns

-25 -15 -5 5 15-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

Severo-Almetyevskaja area

Berezovskaja area

Lag, months

Lag, months Lag, months

Lag, months Lag, months

Lag, months Lag, months

Activity with Injection (detrended) Activity with Production (detrended)

Activity with Injection EffectivenessActivity with Imbalance

> A change of correlation coefficients (between seismic activity and detrended production,injection, imbalance and injection effectiveness) due to shift of data series in time relative to each other. A positive lag, as seen in the cases of the A and M areas, indicates that thechanges in field exploitation parameters precede changes in seismic activity. A negative lagindicates that changes in seismic activity precede changes in exploitation parameters.

Summer 2000 17

Analysis of data on temporal and spatial char-acteristics of seismic activity can provide usefulinformation on the deformation processes occur-ring in reservoirs and surrounding rocks. Zones ofactive faulting that also have high permeability canbe delineated. If acquired over sufficient periods oftime, this information may help forecast hazardousincreases of seismic activity and evaluate recoverymethods. For example, in the Romashkino oil field,water-injection effectiveness decreased duringperiods of increased seismic activity and increasedduring periods of low seismic activity. It may bethat faults that become activated during periods ofseismic activity also develop higher permeability.This could decrease injection effectiveness.

Installing a local permanent seismic networkin advance helps quantify the level of backgroundseismicity so that changes can be detected. Thishelps unravel the mysteries of the relationshipbetween production and seismic activity.Experience shows that to estimate the values oftemporal and spatial parameters of seismicdeformation processes in the region of hydrocar-bon fields, it is advisable to record the data forone or two years in advance of any production.However, more recording and analysis shouldprovide further insight. The results publishedhere are the preliminary findings of the coopera-tive project between Schlumberger and theInstitute of Dynamics of Geospheres in Moscow.Other groups are also actively pursuing surfacemonitoring of seismic activity that may be relatedto hydrocarbon exploitation. For example, theKoninklijk Nederlands Meteorologische Instituut(KNMI) has a program to monitor seismicity inThe Netherlands. Several other groups aremonitoring seismic activity with borehole sen-sors. All of these efforts will improve the indus-try’s understanding of the effects of productionon our surroundings. —LS

Lag, months Lag, months

Lag, months Lag, months

Lag, months Lag, months

Lag, months Lag, months

Activity with Injection (detrended) Activity with Production (detrended)

Activity with Injection EffectivenessActivity with Imbalance

Activity with Injection (detrended) Activity with Production (detrended)

Activity with Injection EffectivenessActivity with Imbalance

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

Almetyevskaja area

Minibayevskaja area

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

-25 -15 -5 5 15 25-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions

-25 -15 -5 5 15 25

-1.0

-0.6

-0.2

0.2

0.6

1.0

Cros

scor

rela

tions